Factor VIII/Von Willebrand Protein MODIFICATION OF ITS CARBOHYDRATE CAUSES REDUCED BINDING TO PLATELETS*

It has been suggested that an abnormality in the carbohydrate structure of the blood clotting glycoprotein, human factor VIII/von Willebrand factor (FVIII/ vWF), may give rise to the hemorrha.gic disorder known as von Willebrand’s disease. More recently, we have reported that the sequential removal of sialic acid and galactose from FVIII/vWF causes a progressive diminution in the ability of human FVIII/vWF to support ristocetin-induced platelet aggregation. We now report experiments aimed at defining how modifications of carbohydrate side chains of FVIII/vWF protein cause a loss of platelet-aggregating activity. The ristocetin cofactor activity of human FVIII/vWF was re- duced to 39% after removal of 74% of the sialic acid by protease-free neuraminidase. Ristocetin cofactor activ- ity was reduced further to 19% after oxidation of 39% of the galactose residues of asialo-FVIII/vWF by galac- tose oxidase treatment and was restored to 33% after potassium borohydride reduction of galactose-oxidized asialo-FVIII/vWF. The receptor-binding potency and affinity of each form of FVIII/vWF derivative was determined by employing a FVIII/vWF receptor-binding assay. The effective concentration to inhibit 50% binding of 0.2

It has been suggested that an abnormality in the carbohydrate structure of the blood clotting glycoprotein, human factor VIII/von Willebrand factor (FVIII/ vWF), may give rise to the hemorrha.gic disorder known as von Willebrand's disease. More recently, we have reported that the sequential removal of sialic acid and galactose from FVIII/vWF causes a progressive diminution in the ability of human FVIII/vWF t o support ristocetin-induced platelet aggregation. We now report experiments aimed at defining how modifications of carbohydrate side chains of FVIII/vWF protein cause a loss of platelet-aggregating activity. The ristocetin cofactor activity of human FVIII/vWF was reduced to 39% after removal of 74% of the sialic acid by protease-free neuraminidase. Ristocetin cofactor activity was reduced further to 19% after oxidation of 39% of the galactose residues of asialo-FVIII/vWF by galactose oxidase treatment and was restored to 33% after potassium borohydride reduction of galactose-oxidized asialo-FVIII/vWF. The receptor-binding potency and affinity of each form of FVIII/vWF derivative was determined by employing a FVIII/vWF receptor-binding assay. The effective concentration to inhibit 50% binding of 0.2 pg/ml of '251-FVIII/~WF to 5 X lo6 platelets and the binding dissociation constant for each form of FVIII/vWF are: 2.0 pg/ml, 1.1 n M for native FWI/ vWF; 14.8 pg/ml, 12.5 nM for asialo-FVIII/vWF; 66 pg/ ml, 53.8 n M for galactose-oxidized asialo-FVIII/vWF; and 30 pg/ml, 18.9 n M , for KBI-L-reduced galactoseoxidized asialo-FVIII/vWF. Furthermore, a linear correlation between the log of receptor-binding affinity and the log of ristocetin cofactor activity was observed. We conclude that the diminished ristocetin cofactor activity of FVIII/vWF having modified carbohydrate side chains results from reduced binding affinity for platelet FVILI/vWF receptors. Our results also indicate that the binding of FVIII/vWF to platelet receptors is functionally relevant with respect to ristocetin cofactor activity.
Factor VIII/von Willebrand factor (FVIII/vWF) is a plasma glycoprotein or glycoprotein complex (-lo6 daltons) with two distinct biological activities, aspects of which were reviewed recently (1). One function is procoagulant activity which corrects the blood coagulation defect in classical hemophilic patients. The other is platelet-aggregating activity which is * This work was supported in part by a research grant from the National Heart, Lung and Blood Institute, National Institutes of Health (HL 15615). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ An Investigator of the Howard Hughes Medical Institute. essential for the formation of hemostatic platelet plugs at the transected end of small vessels. The molecular basis for the two biological activities of this macromolecule remains unresolved. The two biological activities can be dissociated from each other by treatment of FVIII/vWF with thrombin (2) or high ionic strength buffers (3)(4)(5). Some believe that dissociated FVIII procoagulant activity is on a smaller molecule which corresponds to <1% of the FVIII/vWF complex and that the ristocetin cofactor activity is on a much larger molecule which accounts for 99% of the protein in the FVIII/vWF isolate (3-6). Other investigators have suggested that both FVIII and vWF activities are on the same molecule which, after cleavage by thrombin, gives rise to a derivative with full FVIII procoagulant activity (2, 7). Several of the biochemical and immunological characteristics of the FVIII/vWF macromolecule have been described (8, 9). Defects in platelet plug formation and blood coagulation have been observed in patients with a hereditary deficiency of FVIII/vWF protein known as von Willebrand's disease (vWD) (10). With refined laboratory techniques, variant forms of von Willebrand's disease have been described in some patients who have normal levels of plasma FVIII/vWF antigen and FVIII procoagulant activity (11-15). Following a proposal by McKee et al. (16), one group of investigators published findings which suggested that diminution or incompletion of the carbohydrate side chains on the FVIII/vWF protein accounts for its reduced function in some patients with von Willebrand's disease (14). Sodetz et al. reported that successive removal of sialic acid and galactose residues from FVIII/vWF progressively impaired its ristocetin cofactor activity without affecting FVIII procoagulant activity (17, 18); similar results were then observed by Gralnick (19) who removed sialic acid and oxidized the galactose residues of the FVIII/vWF protein.
Recently, we have identified and established the existence of FVIII/vWF receptors on platelets which relate directly to their function in the ristocetin-induced platelet aggregation assay (20, 21). Now, by employing a FVIII/vWF receptorbinding assay, we have further explored and defined the mechanism by which modifications of carbohydrate side chains on normal human FVIII/vWF causes reduced plateletaggregating activity in the presence of ristocetin.

METHODS'
Amerlcdn N a f l o n a l l e d Cross, Retherdd MD R l i t o c e t l n w a s Ourchased from H tundbeck R ~. Mdterlali. lntermedlate-Durlty 'luman FYI:I/vWF concentrate was Obtdlned from the ' Portions of this paper (including "Methods," a small part of " Results," and Figs. 3,4,and 5) are presented in miniprint as prepared by the authors. Miniprint is easily read with the aid of a standard magnifying glass. Full-size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, Md. 20014. Request Document 80M-927, cite author(s), and include a check or money order for $1.20 per set of photocopies. Full sized photocopies are also included in the microfilm edition of. the Journal that is available from Waverly Press. Table I, 74% of the total sialic acid was removed from native FVIII/vWF by neuraminidase treatment. The same quantity of sialic acid was also removed from native FVIII/ vWF during the preparation of galactose-oxidized asialo-FVIII/vWF. After desialylation, the exposed terminal galactose residues were oxidized to galactose aldehyde by treatment with galactose oxidase. Then the residual amount of galactose that had not been oxidized was measured by enzymatic assay for D-galactose. In order to determine whether the enzymatic assay for D-galactose can be used to quantify only galactose in the presence of galactose aldehyde, the substrate specificity of galactose dehydrogenase for galactose aldehyde was studied f i s t . Despite the presence of galactose aldehyde, we found that no more than 3% galactose was detected after galactose (300 nmol/ml) had been incubated for 2 h with galactose oxidase (5 units/ml) and catalase (1 pg/ml) a t 37°C. Therefore, the use of the enzymatic assay is valid for quantitating any galactose remaining in the carbohydrate side chains of galactose-oxidized asialo-FVIII/vWF or KBH4-reduced galactoseoxidized asialo-FVIII/vWF. We found that 39% of the total galactose residues of asialo-FVIII/vWF were oxidized by treatment with galactose oxidase and that the oxidized galactose residues could be restored completely to galactose by borohydride reduction (Table I). Each carbohydrate-modified form of FVIII/vWF (10 pg) was reduced by /I-mercaptoethanol (5%) and analyzed by sodium dodecyl sulfate gel electrophoresis. In each case, only the 200,000-dalton subunit was observed, and as we described before, no evidence of polypeptide chain degradation was seen (18).

Modification of Carbohydrate on FVIII/v WF-As shown in
Platelet Aggregation using Native and Carbohydratemodified FVIII/v WF-The platelet-aggregating activities of native FVIII/vWF and each of its carbohydrate-modified forms were calculated from the reciprocals of their slopes shown in Fig. 1 and are listed in Table I. After removal of sialic acid, the aggregating activity of FVIII/vWF was reduced to 39% of native FVIII/vWF. This activity was reduced further to 19% of native FVIII/vWF after galactose oxidation and was restored to 33.3% of native FVIII/vWF after borohydride reduction. Of interest is our finding that exposure of native FVIII/vWF to the same concentration of potassium borohydride actually decreased its platelet-aggregating activity to -73% of normal. Therefore, the restoration of platelet aggregation activity, in spite of direct adverse effects of potassium borohydride on FVIII/vWF, must indicate that the loss of vWF activity occurred specifically as a result of galactose oxidation. It is most likely that the failure to restore plateletaggregating activity to the exact preoxidation level (39%) is due to the direct effects of KBH, on the protein (25). It should be noted that all the platelet aggregation activities of native or carbohydrate-modifled FVIII/vWF were determined in the same experiment. Moreover, the platelet responsiveness to a given concentration of any FVIII/vWF species tested did not change during the time required for the entire set of analyses.
Receptor-Binding Potency of Carbohydrate-modified forms of FVIII/uwF-The receptor-binding potencies of native or carbohydrate-modified FVIII/vWF were determined from the ability of selected quantities of each FVIII/vWF species to compete with 0.1 pg of l"I"FVIII/~WF for the platelet receptors (Fig. 2). The effective concentration required to inhibit 50% of the specific binding of '251-FVIII/~WF to 5 X lo6 platelets is: 2.0 pg/ml, 14.8 pg/ml, 66 pg/ml, and 30 pg/ml for native FVIII/vWF, asialo-FVIII/vWF, galactoseoxidized asialo-FVIII/vWF, and KBH4-reduced galactose-oxidized asialo-FVIII/vWF, respectively. The parallel relationships observed among the binding competition curves for the native and carbohydrate-modified forms of FVIII/vWF suggest that all compete with '""IFVIII/VWF for the same specific binding sites on the platelet membrane.   " Inhibition constant ( K J determined from Dixon plot; 1.1 X lo6 used for the molecular weight of FVIII/vWF (8).
Inhibition constant (K,) determined from effective concentration which inhibits 50% specific binding (ECso) of 1251-FVIII/~WF (0.2 p g / d ) to 5 X lo6 platelets by using the equation: K, = ECm/l + (L/Kd) where L is the concentration of free 1251-FVIII/~WF and Kd is the dissociation constant of '251-FVIII/~WF binding to platelet receptors.
Rate of platelet aggregation (% transmittance (2') per min for a given concentration of FVIII/vWF or its derivatives). These numbers were determined from the reciprocals of the slopes shown in Fig. 1.

Correlation of Receptor-Binding Affinity and Platelet-aggregating Actiuity-We have reported recently that FVIII/
vWF receptor occupancy is proportional to platelet-aggregating activity (21). Thus, the affinity of each form of FVIII/ vWF for the receptors on platelets should be proportional to its platelet-aggregating activity. The platelet-aggregating activity of each form of FVIII/vWF was calculated from the reciprocal of the slope shown in Fig. 1 FIG. 6. Correlation of ristocetin cofactor activity and receptor-binding affinity for the different forms of FVIII/vWF. The ristocetin cofactor activities of the carbohydrate-modified forms of FVIII/vWF were determined from the reciprocals of the slopes shown in Fig. 1 and then plotted against the receptor-binding affinity of each respective species. The two data points indicated by (xl and x*) were determined from experiments in which FVIII/vWF (xl) and its asialo derivative (xz) were prepared from a different lot of starting FVIII/ vWF concentrate than the data points indicated by 0.
aggregating activity was plotted against the log of receptor binding affinity (KJ for each form of FVIII/vWF (Fig. 6 ) . The linear correlation observed between these two parameters indicates that platelet-aggregating activity of FVIII/vWF is a function of receptor-binding affinity. Therefore, the diminished platelet-aggregating activity of carbohydrate-modified FVIII/vWF is related directly to a decreased binding affinity for the receptors on platelets.

DISCUSSION
We recently established that FVIII/vWF receptors are present on human platelets and that the extent of FVIII/vWF binding to platelets is directly proportional to the extent of ristocetin-induced platelet-aggregating activity (20, 21). Our present results demonstrate that carbohydrate-modifled forms of FYIII/vWF compete with '251-FVIII/~WF for the same binding sites, although at a markedly reduced binding affinity. Significantly; the receptor-binding affinity of each form of FVIII/vWF shows excellent correlation with its platelet-aggregating activity (Fig. 6). Hence, the potency of carbohydrate-modified forms of FVIII/vWF to aggregate platelets is directly proportional to their ability to compete for binding sites on platelets. This observation fulfills an important criterion for establishing that binding measurements truly reflect the interaction of ligand with physiologically relevant receptors (29). Thus, diminished receptor-binding affinity must be responsible for the loss of ristocetin cofactor activity observed after modification of the carbohydrate on FVIII/vWF. In Fig.  1, the reciprocal plot of platelet-aggregating activity versus the concentrations of the different carbohydrate-modified forms of FVIII/vWF suggests that in the presence of ristocetin, the different modified forms of FVIII/vWF induce platelet aggregation by the same mechanism. Hence, the loss of platelet-aggregating activity observed with the carbohydrate-modified FVIII/vWF results from reduced receptor-binding affinity. Whether the reduced receptor-binding a f f i t y of carbohydrate-modified FVIII/vWF is due to conformational change(s) of the molecule, or due to the removal of carbohydrate residues directly involved in receptor binding, remains to be determined. Recently, much attention has been focused on the biological significance of the constituent oligosaccharide units of different glycoproteins (30-34); it has been suggested by some that the carbohydrate side chains may play a role in maintaining the conformational stability of glycoproteins (31)(32)(33). However, it has also been found that removal of sialic acid groups from the (Fab'h fragments of immunoglobulin G abolishes antigen-binding ability, presumably without major conformational change (34). Thus, there are precedents

Binding of Carbohydrate-modified FVIII/v WF to Platelets
with other glycoproteins which would support either a conformational change or a direct alteration in binding affinity when the carbohydrate side chains are modified. The importance of FVIII/vWF carbohydrate side chains for the full expression of ristocetin-induced platelet-aggregating activity has been recognized by biochemical studies of the FVIII/vWF glycoprotein from von Willebrand's disease patients (14) as well as by functional studies of enzymatically modified normal FVIII/vWF (17-19). In the present report, we have confirmed our previous observations (17, 18) that both sialic acid and galactose residues of FVIII/vWF protein are indeed important determinants of ristocetin cofactor activity. When 74% of the sialic acid residues were removed from native FVIII/vWF, we observed a 62% loss of ristocetin cofactor activity (Table I). These values are in good agreement with our earlier report. Recently, we also showed that only 62% of the galactose residues can be cleaved enzymatically from asialo-FVIII/vWF (>95% desialylated) and that this is accompanied by a reduction in ristocetin cofactor activity to -12% of normal (18). In the present experiments, we purposefully performed our desialylation at pH 6.8 since procoagulant activity is not destroyed at this pH, believing this to be a desirable indication that little or no degradation occurred; however, only 74% of the sialic acid could be removed. Hence, 26% of the expected galactose remained unexposed another 38% cannot be exposed despite tryptic digestion of >95% desialylated FVIII/vWF (18). Therefore, it is to be expected that a total of 64% of the galactose residues will not be accessible to galactose oxidase; this predicted value is essentially that which we observed (61%). Importantly, reduction of the oxidized galactose groups restored platelet-aggregating activity almost to that seen for the asialo-FVIII/vWF. These observations clearly extend our earlier results by showing that the reduced ristocetin cofactor activity seen with either asialo-FVIII/vWF or galactose-oxidized asialo-FVIII/vWF is due to proportionate reductions in binding affinity for the platelet receptors.
Gralnick (19) disagrees that removal of sialic acid from FVIII/vWF causes a loss of ristocetin cofactor activity. There are several possible reasons for the discord between his results and ours. For example, paraformaldehyde-fiied platelets were used by Gralnick (19) whereas we have always used freshly washed human platelets. Moreover, our data indicate that reduced ristocetin cofactor activity observed with the carbohydrate-modified forms of FVIII/vWF becomes most obvious when the assay is performed over a range of low-concentrations of the particular FVIII/vWF species (Fig. 1). Therefore, unless one is aware that the assay sensitivity is amplified for detecting differences in ristocetin cofactor activity over the lower concentration range, the decreased ability of asialo-FVIII/vWF to support ristocetin-induced platelet-aggregating activity might be missed. Lastly, but of equal importance, whether the decrement in ristocetin cofactor activity is detected when sialic acid is removed from native FVIII depends in part on the initial ristocetin cofactor activity of the purified, native FVIII/vWF used as starting material. We fiid that normal human FVIII/vWF preparations purified from different lots of starting concentrates have different amounts of ristocetin cofactor activity. However, when the sialic acid is removed from any of these purified FVIII/vWF preparations, all have about the same level of ristocetin cofactor activity, albeit reduced in varying proportion from that observed for the starting material, An example of this phenomenon is given in Fig. 3. For this reason, purified native FVIII/vWF with a low starting ristocetin cofactor activity may make it difficult to detect significant reductions in ristocetin cofactor activity after desialylation.
Our results continue to support plausible models for the two most frequently encountered forms of von Willebrand's disease. In the instance of asialo-FVIII/vWF, low plasma levels of the antigen would be expected because of rapid clearance by the hepatic receptor which recognizes asialoglycoproteins; this we established in the past (17). Although continuing to possess full FVIII procoagulant activity, the decreased sialic acid on the asialo-FVIII/vWF molecule also contributes to decreased ristocetin cofactor activity by virtue of its reduced affinity for its receptor on the platelet.
Our findings with agalacto-FVIII/vWF suggest analogies to those features of the less frequently encountered form of von Willerbrand's disease in which normal plasma levels of FVIII/ vWF antigen and FVIII procoagulant activity are observed (11-15). For example, the agalacto-FVIII/vWF has normal FVIII procoagulant activity, no affinity for the hepatic receptor (suggesting a normal circulatory survival and therefore a normal antigen level), but a markedly reduced ristocetin cofactor activity (18). Our present study expands the latter finding by showing that galactose-oxidized FVIII/vWF has only about 20% of the expected affinity for the FVIII/vWF platelet receptor and this correlates well with its decreased function in the ristocetin cofactor activity assay. Albeit neither of these models has been proven, we do suggest that our findings may in part explain or provide leads to the eventual definition of the molecular defects responsible for the spectrum of von Willebrand's disease.